Javier Orduz retweeted
The attention mechanism, including the K, Q, V, never made much sense to me. Until I started seeing it as a word gravity, where words influence each other the way planets pull each other.
If you have two minutes, learn about it here!
Javier Orduz retweeted
Four matrix factorisations cover most of numerical linear algebra: LU, QR, the singular value decomposition and Cholesky. Each rewrites a matrix as a product of simpler, structured pieces.
LU with pivoting, PA = LU, is Gaussian elimination recorded as matrices and solves square systems Ax = b. QR splits A into an orthogonal Q and an upper triangular R, the stable route to least squares. The SVD, A = UΣVᵀ, exists for every real matrix and exposes its rank and principal directions. Cholesky, A = LLᵀ, applies only to symmetric positive definite matrices and costs about half as much as LU.
Truncating the SVD to its largest singular values gives the best low-rank approximation of a matrix, which is the basis of PCA and many compression methods.
Javier Orduz retweeted
Why does the bottom person always carry more of the sofa?
As the incline steepens the center of gravity shifts relative to the supports. The upper person mainly counters torque with a force nearly perpendicular to the object while the lower person balances weight and residual components. Bottom load rises from half toward nearly all.
This lever principle shapes robotic grasps and structural paths. Geometry sets the effort before strength does.
Javier Orduz retweeted
Updated 2236-page PDF Mathematics ebook…
"Algebra, Topology, Differential Calculus, and Optimization Theory For Computer Science and Machine Learning"
Find it here: cis.upenn.edu/~jean/gbooks/g…
Javier Orduz retweeted
One symbol decides the sign of every ordered triple in three dimensions.
ε_ijk equals +1 when (i,j,k) is an even permutation of (1,2,3), −1 for an odd permutation, and 0 the moment any index repeats. That single rule is all it takes.
The same numbers assemble the determinant of a matrix as a signed sum of products and force the cross product a × b to stand exactly perpendicular to both vectors.
Orientation is nothing more than the memory of which order we agreed to call positive.
Javier Orduz retweeted
Download this incredibly comprehensive 1048-page PDF eBook…
Mathematics for Computer Science: courses.csail.mit.edu/6.042/… from @MIT @MIT_CSAIL
—————
#ComputerScience #Algorithms #ComputationalScience
Javier Orduz retweeted
Foundations of Large Language Models: arxiv.org/abs/2501.09223
[download 315-page PDF]
Javier Orduz retweeted
The Little Book of Generative AI Foundations — Intuitive Mathematical Primer: arxiv.org/pdf/2605.29713
[download 195-page PDF]
Javier Orduz retweeted
Download 348-page PDF eBook >> Machine Learning for Scientists and Engineers: smlbook.org/book/sml-book-dr…
————
#DataScience #DataScientist #ML #Algorithms #Mathematics
Javier Orduz retweeted
“Introduction to Applied Linear Algebra: Vectors, Matrices, and Least Squares”
Download 473-page PDF here: web.stanford.edu/~boyd/vmls/…
Javier Orduz retweeted
#MartesDeVacantes
💼Te estamos buscando. Forma parte del equipo de #ONUMéxico
Consulta más en: 📌bit.ly/3ciBxEX #SomosONU
Javier Orduz retweeted
Hay personas cuyo legado está en la ciencia y ni siquiera participaron en ella durante la vida, sólo después de la muerte. ¿Tú te atreverías a donar el cuerpo que ya no usarás para la preparación de los médicos?
Javier Orduz retweeted
A 2×2 matrix A makes the plane flow by x′ = Ax, and each flow is one of nine kinds: points sink, escape, spiral, circle or settle on a line.
The kind is set by the eigenvalues of A, the roots of λ² − (tr A)λ + det A = 0. det A < 0 gives a saddle. Above det A = (tr A)²/4 they are complex: spirals, or a centre if tr A = 0. Below it, nodes; on it, star and degenerate node.
The stable spiral here has tr A = −1, det A ≈ 2.65, so λ = −0.5 ± 1.55i.
Javier Orduz retweeted
Mathematics begins with mere collections of objects, yet builds towering castles of abstract symmetry.
We start with a simple Set equipped with a binary operation, stepping up into Monoids and Groups where every element finds its inverse.
As we introduce a second operation, these structures blossom into Rings and Fields, capturing the familiar arithmetic of numbers.
Javier Orduz retweeted
An eigenvector of a matrix is a direction it does not turn, only scales: Av = λv.
Here A is blended in from the identity, (1 − s)I + sA. The box on (−1, 2, 2), (2, −1, 2) and (2, 2, −1) never turns; its edges only scale, by 1.8, 1.2 and 0.6, the roots of det(A − λI) = 0, and a point inside rides along to Ax.
A is symmetric, so the three directions are perpendicular, and A = λ₁v₁v₁ᵀ + λ₂v₂v₂ᵀ + λ₃v₃v₃ᵀ for the unit vᵢ.
Javier Orduz retweeted
Foundations of Large Language Models: arxiv.org/abs/2501.09223
[download 315-page PDF]
Javier Orduz retweeted
Updated 2236-page PDF Mathematics ebook:
"Algebra, Topology, Differential Calculus, and Optimization Theory For Computer Science and Machine Learning"
Find it here: cis.upenn.edu/~jean/gbooks/g…
Javier Orduz retweeted
The upside-down triangle in every physics textbook started life on its side.
Hamilton wrote it as a sideways ▷ in 1846. Tait turned it into ∇ in 1867; Maxwell, Gibbs and Heaviside made it the language of fields.
Today it's one operator: ∇ = x̂ ∂/∂x + ŷ ∂/∂y + ẑ ∂/∂z
On a scalar, ∇f points uphill fastest. Dotted with a field, ∇·F measures outflow. Crossed, ∇×F measures rotation.
One symbol, three questions you can ask any field.
Javier Orduz retweeted
University of Toronto's "Neural Networks for Machine Learning"
by Geoffrey Hinton
Videos: cs.toronto.edu/~hinton/cours…
Slides: cs.toronto.edu/~hinton/cours…
Javier Orduz retweeted
Free Machine Learning PDF: Pen and Paper Exercises in Machine Learning
Want to strengthen your Machine Learning fundamentals without relying entirely on code?
Pen and Paper Exercises in Machine Learning is a learning resource for practicing ML concepts through mathematical reasoning and step-by-step problem-solving.
What you can explore:
Machine Learning concepts and mathematical foundations
Problem-solving through pen-and-paper exercises
Analytical thinking for ML algorithms
Practice to strengthen your understanding of ML theory
PDF Length: 211 Pages
Resource: Free PDF
Read and access the resource:
clcoding.com/2026/08/pen-and…





























